Diagnosability of Input Output Symbolic Transition Systems

Diagnosability checking of discrete-event systems hasbeen extensively studied in the framework of classical nonsymbolic models such as Labeled Transition Systems. Ithappens that in practice such models tend to need too muchspace to be efficiently processed. By opposition, symbolic approachesoffer an expressive, easy and concise way to modelsystems, and checking diagnosability from such symbolicmodels can benefit from this reduction of space complexity.Indeed, though this will generally translate into time complexity,such a tradeoff is advantageous, as diagnosabilitychecking is something that is usually done at design stage.This is why this paper proposes a theoretical frameworkto check diagnosability of Input Output Symbolic TransitionSystems (IOSTS) by adapting the twin plant approach to thesymbolic case and relying on the use of a symbolic modelchecker. This theoretical work is being currently applied toembedded functions inside a vehicle in the context of anindustrial project and a simplified version of this problemwill serve as a running example throughout the presentation.